Independent public-signal monitor

See the public signal
before you check your quota.

A quiet, evidence-first radar for Codex usage resets. See the public signal, its source time, and when it was first observed—without mistaking a fresh collector check for a fresh source event.

Three clocks stay separate: source published time, first collector observation, and your own account result.

60s collector cadenceAwaiting first primary-source timingSHA-256 public receiptsScope + type on every recordAccount outcomes stay separate

Check the latest evidence Get one confirmed alert

No Codex login · UTC timestamps · local verification

Fresh collector check ≠ fresh source event.

Collector Checking

Last confirmed reset

Aug 13, 2026

Aug 13, 2026, 1:01 AM UTC

Evidence clock: UTC · Display: UTC

Newer context signal: Sep 2, 2026, 5:07 PM UTC

Record confidence88%

Confidence in this public record · not account-level accuracy

1 primary0 corroborating1 context

Claim scope: broad public · Timing: historical, outside current window

View live evidence boundary

verified
public signal

updated
Sep 3, 2026, 2:44 PM UTC

66tracked signals

8.7 dayshistorical gap

60scollector cadence

1 emailper confirmed public signal

Choose the right layer

One question, three
useful paths.

ResetPulse is not a replacement for your private quota view. Pick the job you actually need, then use the smallest surface that answers it.

Did a public reset happen?

Read the latest signal, UTC source and observation times, scope, event type, and the receipt another person can verify.

Open the public answer

What does my account show?

Use your Codex client as the account-level source of truth, then compare a sanitized before-and-after snapshot locally without uploading credentials.

Check privately in this browser

Can I bring the signal into my work?

Add a credential-free badge, verified packet, RSS feed, or user-owned notification workflow to a README, dashboard, or community tool.

Open the two-minute quickstart

Evidence lane health

One check is not
one source.

ResetPulse shows which public lanes were reachable, how many items qualified, and when a lane had no claim to make. This is the operational evidence behind the timeline—not a hidden speed slogan.

Primary public lanechecking

HTTP — · —ms · accepted public records

Direct profile lanechecking

HTTP — · —ms · claim state pending

Parallel X relaychecking

HTTP — · —ms · same-origin relay, not independent proof

RSS fallback relaychecking

HTTP — · —ms · same-origin fallback, not independent proof

Reddit corroborationchecking

HTTP — · —ms · corroboration never confirms alone

Open the live health receipt · All timestamps are UTC; a reachable lane is not proof that every account changed.

The layer we own

Not another quota panel.
A public proof layer.

Personal trackers answer “what does my account show?” ResetPulse answers “what happened publicly, how fast did we observe it, and can I verify the claim?”

Race public source lanes

Direct public-source and archive checks run in parallel while your computer is off; the receipt keeps the earliest observed time instead of hiding archive lag.

Classify before sharing

Every event carries an evidence state, claim scope, and event type so a full refresh is not confused with a banked credit or an announcement.

Make proof portable

One event becomes a readable page, JSON receipt, RSS item, schema, badge, and shareable evidence card for communities and developer tools.

Show propagation

A shareable trace separates source time, first collection, evidence state, and anonymous account or plan outcomes instead of collapsing them into one counter.

Create a fresh proof update · Build with the public signal · Connect your local outcome · See the honest comparison · Read the evidence model

A more useful radar

Forecast the public signal.
Plan your own window.

Competitors offer forecasts or personal timers. ResetPulse now keeps both, but makes the boundary explicit: a public estimate never pretends to be your private quota.

Transparent 24–48h baseline

30%within 24 hours

35%within 48 hours

Median gap: 2.3 days14 historical gaps0 gaps still at riskConfidence: lowBacktest: 58% / 50% (12 trials)Brier: 0.296 / 0.329 (lower is better)

Elapsed-aware empirical estimate from deduplicated broad-public full-refresh gaps, shrunk toward a three-gap short-horizon baseline when the at-risk sample is small; not a personal-account forecast. Backtest accuracy and Brier score are descriptive, not a guarantee or a calibrated promise. They describe a broad public signal, not an account-specific reset.

Private personal planner

Keep your account clock separate.

Pair the remaining usage percentage with the reset time shown by your own Codex client. The public signal stays beside it for context, while your account data stays in this browser.

Remaining usage (%)Personal weekly reset timeBanked reset grant date

Remaining usage: Not setPersonal window: Not setBanked expiry: Not set

Public signalConfirmed reset

Private usageNot set

BoundaryNot inferred

Browser reminder

Local only: this works while this page remains open on this device. It never reads or stores your Codex login. For reminders that work with the page closed, use calendar export or your own notification channel.

Community pulse

Help separate the signal
from account reality.

Report only what your own Codex surface showed. No login, email, free text, credentials, or account identifier is collected. Reports stay unverified and are shown publicly only as an aggregate sample.

Event-level anonymous report

Did this public signal reach your surface?

Report only what your Codex surface showed for the latest public signal. Use UTC-normalized time and choose the closest scope. A community report can reveal rollout differences, but it can never confirm a public reset on its own.

Observation timeSurfaceScope

The event-level sample appears after three reports; individual observations stay private.

The signal log

A clear history
of public signals.

Every entry separates source time, observation time, and confirmation state. No forecast is presented as proof. No guesswork. No noise.